In case it isn't clear in the markdown preview, the agent has access to honeybadger-mcp-server tools, which allows it to look up info about production errors in Honeybadger:
docs.honeybadger.io/resources/ll...
Look at the raw markdown frontmatter to see how it works.
That looks awesome! Production error -> LLM pull request seems like a winning pattern
Josh from Honeybadger
Thanks! That’s the idea. We also already auto create GH issues from errors, so I think you could just assign the agent.
Depends on the error and how good the model is of course but should definitely handle some common issues.
Peter Van Dijck
Or production error -> LLM hypothesis of the issue root cause
Peter Van Dijck
Another workflow: honeybadger: “I went through the code, the root issue is likely x, a few ways to solve this may be a and b.” Copy and paste into Claude.